High-Resolution Melting Analysis to Detect Antimicrobial Resistance Determinants in South African Neisseria gonorrhoeae Clinical Isolates and Specimens

Background Antimicrobial resistance is limiting treatment options for Neisseria gonorrhoeae infections. To aid or replace culture and the syndromic management approach, molecular assays are required for antimicrobial susceptibility testing to guide appropriate and rapid treatment. Objective We aimed to detect single-nucleotide polymorphisms and plasmids associated with antimicrobial resistance from N. gonorrhoeae isolates from a clinic population in South Africa, using real-time PCR as a rapid test for AMR detection. Methods N. gonorrhoeae isolates, from female and male patients presenting for care at a sexually transmitted infections clinic in Durban, South Africa, were analysed using phenotypic and genotypic methods for identification and antibiotic susceptibility testing (AST). Real-time PCR and high-resolution melting analysis were used to detect porA pseudogene (species-specific marker) and resistance-associated targets. Whole-genome sequencing was used as the gold standard for the presence of point mutations. Results The real-time porA pseudogene assay identified all N. gonorrhoeae-positive isolates and specimens. Concordance between molecular detection (real-time PCR and HRM) and resistance phenotype was ≥92% for blaTEM (HLR penicillin), rpsJ_V57M (tetracycline), tetM (tetracycline), and gyrA_S91F (ciprofloxacin). Resistance determinants 16SrRNA_C1192U (spectinomycin), mtrR_G45D (azithromycin), and penA_D545S, penA_mosaic (cefixime/ceftriaxone) correlated with the WHO control isolates. Conclusions Eight resistance-associated targets correlated with phenotypic culture results. The porA pseudogene reliably detected N. gonorrhoeae. Larger cohorts are required to validate the utility of these targets as a convenient culture-free diagnostic tool, to guide STI management in a South African population.


Introduction
Sexually transmitted infections (STIs) are a global public health concern, with an annual estimate of 87 million new cases of Neisseria gonorrhoeae infection [1]. Globally, N. gonorrhoeae has developed resistance to most antibiotics, including third-generation cephalosporins, cefixime and ceftriaxone. e World Health Organization (WHO) call to end the STI epidemic as a public health concern emphasizes a need to strengthen technologies and improve knowledge on the prevalence, resistance patterns, and public health interventions to combat STIs, especially in low-and middle-income countries [2].
For N. gonorrhoeae, the sensitivity of microscopy for diagnosis is 90-95% in men and 50-70% for presumptive diagnosis in women [3]. e sensitivity of culture ranges from 85-95% in patients with recent and symptomatic infection to as low as 50% in asymptomatic patients [4,5]. A challenge to STI management is delayed or ineffective treatment initiation whilst awaiting laboratory diagnosis (48-72 hours), which could lead to patient loss to followup, [6,7] and increased risk of infection transmission to sexual partners. In well-resourced settings, sensitive and rapid nucleic acid amplification tests (NAATs) have largely replaced phenotypic identification of N. gonorrhoeae [5]. Although NAATs are effective in diagnosis at POC, most settings treat patients empirically at the first visit.
Syndromic and empiric treatment contributes to the development of resistance to currently recommended drugs in many parts of the world [8][9][10]. N. gonorrhoeae mechanisms of resistance have been well documented [11][12][13].
ese include plasmid-mediated resistance to penicillin (bla TEM ) and tetracycline (tetM) and chromosomally mediated resistance to penicillin, tetracycline, spectinomycin, fluoroquinolones, azithromycin and cephalosporins [11]. Resistance to penicillin and extended-spectrum cephalosporins has been associated with modifications and recombination within penA, porB, and ponA [14]. e mutation L421P in ponA reduces the rate of acylation with penicillin [15]. Mutations in mtrR, as well as its promoter region, can cause overexpression of the mtrCDE efflux pump which has been associated with resistance to penicillin, tetracycline, cefixime, ceftriaxone, and azithromycin [16,17]. Mutations in porB which encode porinB, reduces the porin permeability. e pilQ gene encodes for pore formation in the outer membrane, mutations in which result in reduced antibiotic influx [11,13,18,19]. Resistance to tetracycline has been associated with mutations in rpsJ, mtrR, and porB [16]. e rpsJ mutation V57M alters the binding site, thus reducing the binding affinity of tetracycline for the ribosome [11]. Resistance to ciprofloxacin is due to mutations in gyrA and parC [16]. High-level resistance to spectinomycin is due to the mutation C1192U in 16S rRNA by reducing antibiotic target affinity [20]. Resistance to azithromycin is often due to mutations in 23S rRNA, namely, C2611T (low-level resistance) or A2059 (high-level resistance) [11,20].
Molecular antimicrobial resistance testing is imperative for the control of N. gonorrhoeae, to facilitate antibiotic stewardship, to expedite appropriate treatment of patients, and to conserve the effectiveness of the current treatment regimen [21]. A recent study projected that continued empiric treatment without antimicrobial susceptibility testing at the POC would result in >5% of N. gonorrhoeae isolates being resistant to both azithromycin and ceftriaxone within 15 years, but the use of a molecular assay could delay the emergence of resistance by 5 years [22,23]. Currently, the only commercially available genotypic resistance testing assay for N. gonorrhoeae is from SpeeDX (Australia), which determines resistance to ciprofloxacin by detecting the gyrA_S91F mutation. However, due to the high prevalence of resistance to ciprofloxacin in South Africa and other African countries [21], additional assays to detect genotypic resistance to other drugs are necessary.
Whole-genome sequencing and a variety of bioinformatics tools are widely used to predict AMR and identify novel mutations in N. gonorrhoeae. ese tools include Resistance Gene Identifier (RGI), Basic Local Alignment Search Tool (BLAST), Rapid Annotation Using Subsystems Technology (RAST), Antibiotic Resistance Gene-ANNOTation (ARGANNOT), ResFinder, ARIBA, ABRicate, ResFinder, PointFinder, and NG-STAR [24]. However, the cost of this technology is high, especially in low-middle-income-countries (LMIC).
Real-time polymerase chain reaction (real-time PCR) combines amplification and detection of gene targets in a single assay. Advantages include rapid detection, the ability to be implemented in high throughput settings [25], and cost effectiveness [26]. e porA pseudogene is a N. gonorrhoeae species-specific marker [27][28][29] and is highly conserved across a diverse range of N. gonorrhoeae strains making it a useful target for identification [27,28,30,31].
High-resolution melt (HRM) analysis is a post-PCR analysis method, which amplifies gene targets in the presence of a fluorescent reporter dye; the increasing temperature gradient denatures the DNA and releases the dye, which then undergoes a conformational change that reduces the fluorescence. An HRM instrument records the fluorescence fluctuation and produces a melt curve of the target gene, which can be used to compare the similarity of PCR products [32]. HRM is so sensitive that it can detect a singlenucleotide polymorphism (SNP) [30].
is technology is rapid and cost effective [33], a cheaper alternative to sequencing [32,34], and can be employed to detect SNPs associated with antibiotic resistance [30,35,36].
Developing and evaluating new technologies for AMR prediction directly from patient specimens, to be used at the POC or near-patient, is currently a priority to conserve current and future antimicrobials [18]. Some resistance determinants may work on their own in increasing resistance in particular drugs. However, in some cases, a single determinant may only alter the minimum inhibitory concentration (MIC) slightly or not at all [11]. In our study, we correlated known AMR determinants with phenotypic AST (gold standard) data from our local isolates to determine which mutations in our population result in a resistant phenotype. Using real-time PCR and HRM technologies, we detected genes and mutations associated with antimicrobial resistance to penicillin, tetracycline, spectinomycin, ciprofloxacin, azithromycin, ceftriaxone, and cefixime in our local population.

Isolation and Identification of Neisseria gonorrhoeae.
Phenotypic identification included bright-field (Gram stain) microscopy and culture. Swabs were inoculated onto New York City and chocolate agar media and incubated for 24-48 hours at 37°C in a 5% CO 2 incubator. Suspected N. gonorrhoeae colonies were confirmed using the Rapid Oxidase test and Phadebact ® Monoclonal GC test (Pharmacia, Sweden). STI screening, directly from patient specimens, was performed using NAATs, Anyplex ™ II STI-7 Detection (Seegene, Seoul, Korea), and Xpert® (Cepheid, CA, USA). Isolates were stored in the laboratory repository for future phenotypic and genotypic analysis.

Antimicrobial Susceptibility Testing. N. gonorrhoeae
isolates from vaginal and urethral specimens were revived on nonselective ayer Martin (antibiotic supplement excluded) and chocolate agar media for 18-24 hours in a 37°C 5% CO 2 incubator. WHO N. gonorrhoeae control strains (F, G, K, L, M, N, O, and P) [38] and ATCC strain 49226 were used in this study. AST using Etest ® (bioMérieux, Marcy l'Etoile, France) was performed for penicillin, tetracycline, spectinomycin, ciprofloxacin, azithromycin, ceftriaxone, and cefixime, as per the manufacturer's guidelines, using GC agar base medium supplemented with 1% Vitox [39,40]. e European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines [41] were used to interpret MICs. 2.5. Real-Time PCR for porA, bla TEM , and tetM. Real-time PCR was performed on the Quant Studio 5 ( ermoFisher, CA, USA), to detect porA for identification of N. gonorrhoeae and plasmids bla TEM and tetM, which confers resistance to penicillin and tetracycline, respectively.

Statistical
Analysis. e sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for HRM in comparison with WGS and phenotypic AST were calculated using MedCalc calculator (MedCalc Software bvba, Ostend, Belgium). Sequencing data were used as the reference for the identification of resistance-associated genes and mutations. Phenotypic AST was used as the reference for resistance prediction.

Genotypic Analysis.
A total of 22 paired N. gonorrhoeae isolates and clinical specimens and eight WHO control strains were included in this study. Real-time PCR detected the porA pseudogene, to identify N. gonorrhoeae, in all 22 (100%) clinical isolates and specimens, and eight control strains. We observed that the bla TEM plasmid showed 100% concordance in the detection of high-level resistance to penicillin when phenotypic and genotypic results were compared. Wholegenome sequencing confirmed the presence of tetM and showed 100% concordance with the phenotypic data.
When evaluating the HRM assay to screen the 22 isolates paired clinical specimens (vaginal/urethral swabs) for the presence or absence of resistance-conferring mutations, the concordance of HRM detection between patient isolate and specimen was >90% for all targets (Table 4). ese excellent concordance values between patient isolate and paired specimen suggest that the data from Tables 2 and 3 can be inferred to patient specimens.

Discussion
is study used molecular techniques to identify N. gonorrhoeae and to detect genes and mutations associated with antimicrobial resistance. AST revealed that N. gonorrhoeae resistance to penicillin, tetracycline, and ciprofloxacin was high in our isolates. We report that spectinomycin, cefixime, and the drugs used in the syndromic management approach, ceftriaxone and azithromycin, remain effective as all isolates were susceptible at the lowest MIC. Our study shows that the sensitivity and specificity of the molecular assays for bla TEM (high-level resistance to penicillin), rpsJ_V57M (lower-level resistance to tetracycline), tetM (high-level resistance to tetracycline), and gyrA_S91F (resistance to ciprofloxacin) were 100%, with 100% concordance to phenotypic ASTdata of patient isolates and control strains. Head-to-head comparison of HRM data International Journal of Microbiology from isolates and paired clinical specimens showed >90% concordance, which indicates the potential to identify AMR targets directly from N. gonorrhoeae-positive vaginal/urethral specimens. While multiple commercially available NAATs are available to diagnose STIs, this study used real-time PCR to detect the species-specific porA pseudogene that identifies N. gonorrhoeae. e concordance was 100% when compared with WGS. Other studies have also reported excellent clinical performance of porA [12,27,30,[48][49][50]. ere have been reports of porA mutants which result in a false-negative for identification [51]; however, this was not detected in our isolates. is cost-effective real-time assay, as a diagnostic marker for reliable identification, can be introduced in a clinic management algorithm, and a result can be obtained within two hours compared to culture, which can take up to 48 hours for isolation of N. gonorrhoeae. e prevalence of the bla TEM (high-level resistance) in our study was 79%. is is similar to a recent study from Africa, which reported a prevalence of 72% [52]. Resistance in isolates with intermediate MICs was chromosomally mediated. Modelling analysis identified tetM and rpsJ_V57M to be excellent predictors of resistance to tetracycline [53]. All isolates in this study were resistant to tetracycline. e prevalence of the tetM plasmid (high-level resistance) was 95%. is is consistent with that of Kenya and South Africa reported recently as 86% and 92%, respectively [52,54]. rpsJ_V57M (lower-level resistance) was also detected in all isolates resistant to tetracycline. is is similar to that reported in Vietnam [55], but higher compared to that found in Johannesburg, South Africa (70%) [56], Brazil (39%) [57], and Ukraine (67%) [58].
Resistance to ciprofloxacin was detected in 82% of isolates, and gyrA_S91F SNP was detected in all resistant isolates. Other studies have also shown that this SNP is an excellent target for ciprofloxacin resistance detection [55,56]. e only available commercial assay for resistance detection, SpeeDX, has been developed based on this mutation.
All isolates from our patient population were susceptible to spectinomycin, and using HRM, we detected the resistance-associated SNP C1192U in the control strain. Spectinomycin resistance is exceedingly rare globally [59], and based on local resistance data, it is an effective treatment for genital and anorectal gonococcal infection [60]. A combination of ceftriaxone and spectinomycin is currently used in Japan [11]. South Korea also effectively treats gonorrhoea with spectinomycin, and resistance has not been reported since 1993 [61]. It is reassuring to know that we have the option to preserve the effectiveness of azithromycin and cefixime/ceftriaxone for use in gonorrhoeal infection by using spectinomycin as first-line therapy.
A limitation of this study is the small sample size, which has resulted in the low performance characteristics of some of the SNP targets when comparing HRM and WGS data. To better determine the performance, further evaluation of these drug targets is required on a much larger sample set from clinical specimens collected from hospitals and clinics across the different provinces in South Africa. porA mutants that result in false-negative identification for N. gonorrhoeae have been found in other studies [51], and in our subsequent studies, a combination of porA and opa will be used for identification. Also, the principal gene for azithromycin resistance, 23S rRNA, was not included in this HRM study and will be included in subsequent studies. However, from WGS data analysis, none of the isolates had mutations present in their 23S rRNA genes. At present, due to unknown and novel mechanisms of resistance, genotypic AMR prediction cannot completely replace phenotypic AMR; therefore, continued and updated surveillance of local isolates is needed to identify mutations associated with resistance so that local in-house assays can be updated accordingly.
Cost-effective molecular diagnostic tools are required for rapid AMR detection, especially in low-middle-income countries. Using a diagnostic tool such as HRM, which is affordable and where multiple targets can be run in a single analysis, is an option for use in antimicrobial stewardship, International Journal of Microbiology rather than the currently used syndromic management approach. However, a functioning laboratory and trained personnel are required. We show that, in our local setting, the porA pseudogene can be used reliably to detect N. gonorrhoeae, bla TEM can be used to detect HLR to penicillin, tetM and rpsJ can be used to detect resistance to tetracycline, and gyrA_S91F can be used to detect resistance to ciprofloxacin.

Data Availability
Whole-genome sequence data are available in DDBJ/ENA/ GenBank with the BioProject number PRJNA681740.
Disclosure e funders had no role in study design, data collection and interpretation, or the decision to submit for publication.

Conflicts of Interest
e authors have no conflicts of interest.